Primary progressive aphasia (PPA) is a clinical syndrome in which degeneration of language regions in the dominant hemisphere is associated with progressive deficits in speech and/or language function. The overall goals of this project are to use functional magnetic resonance imaging (fMRI) to investigate neural changes underlying linguistic deficits in PPA, and to use this information to better discriminate patients with variants of PPA from each other and from normal aging. Recent studies have identified three clinical variants of PPA: progressive non-fluent aphasia (PNFA), semantic dementia (SD) and logopenic progressive aphasia (LPA). Each variant is associated with characteristic linguistic features, distinct patterns of brain atrophy, and different likelihoods of particular underlying pathogenic processes, making correct differential diagnosis highly relevant. We will recruit 48 patients with PPA (16 of each variant) and 24 normal controls over a three year period, and acquire fMRI data along with structural MRI, linguistic and cognitive measures. The fMRI paradigm consists of a syntactic processing task with seven conditions parametrically varying in syntactic complexity. The research will address two specific aims. The first is to identify the relationships between volume loss, changes in functional MRI activation, and linguistic deficits, in the different PPA variants. The second aim is to improve differential diagnosis of PPA variants using machine learning algorithms incorporating both structural and functional imaging measures.